ALMA Antenna responses in CASA imaging

Transcription

1 ALMA Antenna responses in CASA imaging Dirk Petry (ESO), December 2012 Outline Motivation ALBiUS/ESO work on CASA responses infrastructure and ALMA beam library First test results 1

2 Motivation ALMA covers a wide frequency range with 10, possibly 11 different receivers. ALMA performance (τ=60s, 50 antennas) ALMA data can only be taken in one band at a time Bands 3, 6, 7, and 9 have been available for Early Science since Bands 4, 5, 8, and 10 have seen first light. Bands 1, 2 and 11 in preparation. 2

3 Motivation FOV often too small to cover extended objects - i.e. imaging will often be mosaicing, sometimes wide-field imaging' - direction-dependent effects need to be taken into account more accurately - requires good knowledge of the instrument's primary beam (dropping assumptions like rotational symmetry) To achieve high continuum sensitivity, need to integrate over wide band - frequency-dependence of the primary beam becomes important (although less so than for EVLA since fractional bandwidth smaller) - the source spectrum may need to be unfolded in parallel (often have thermal and non-thermal sources in the same image) - 4 different antenna types: Main Array: DA (= AEM), DV (= Vertex), PM (=Melco) all 12 m but different struts ACA: 7m antennas Full use of ALMA or ALMA + ACA requires handling of heterogeneous baselines ALBiUS funded work on ALMA primary beams at ESO (work continues, funded by ESO) 3

4 ALMA/CASA imaging work at ESO a) Assessment of Status Quo Up to CASA 3.2 (mid 2011), all primary beams in CASA are analytical For ALMA it is an Airy disk. Band 3 observation of a 11 point mosaic on a bright quasar with 12 ALMA antennas in May 2011 analysed with CASA 3.2 immediately demonstrated the strong need for improvements. relative reconstr. flux 1.6 * * * * * * * * * * * mosaic pattern (gives 3 offsets) Adjusting the ALMA Airy disk Green: 12.0 m Airy disk (v3.2) 1.4 Red: m Yellow: 10.4 m 1.0 Blue: m 43 arcsecs offset 4

5 ALMA/CASA imaging work at ESO a) Assessment of Status Quo Up to CASA 3.2 (mid 2011), all primary beams in CASA are analytical For ALMA it is an Airy disk. Band 3 observation of a 11 point mosaic on a bright quasar with 12 ALMA antennas in May 2011 analysed with CASA 3.2 immediately demonstrated the strong need for improvements. relative reconstr. flux 1.6 * * * * * * * * * * * mosaic pattern (gives 3 offsets) Adjusting the ALMA Airy disk Green: 12.0 m Airy disk (v3.2) Still inaccurate by 10% at larger offsets 1.4 Red: 10.7 m adopted in v3.4 (in prerelease Dec. 11) Yellow: 10.4 m Blue: m 43 arcsecs offset 5

6 CASA status since v3.4 Primary beam library available for actual imaging: EVLA, ALMA, ATA, ATCA, BIMA, HATCREEK, CARMA, GBT, GMRT, IRAMPDB, SMA, WSRT ALMA, EVLA, ATCA, GBT, and WSRT (and SMA?) are polynomials or Airy disks fitted to measurements of the beam of the actual telescope. Only VLA includes beam squint. In particular ALMA uses an Airy disk for an effective dish diameter of 10.7 m (i.e. scaled down from 12 m by factor 1.12) E.g., the Fomalhaut ApJ Letter (Boley et al., 2012) used this PB down to 7% power. 6

7 b) enable use of response images 1) Creation of the AntennaResponses C++ class in casacore - a database for antenna responses in different representations - define which primary beam is in use for which observatory which antenna type which time range which elevation range which beam (if there is more than one) 2) Extension of the BeamCalc ray tracing code (W. Brisken) for general use + adding astigmatism feature 3) Upgrade of the Voltage Pattern Manager ( vpmanager ) to connect to AntennaResponses databases - user interface at the Python level of CASA 4) (ongoing) Tests and optimisations of the above 7

8 b) enable use of response images Best source of ALMA responses so far: The TICRA simulations of ALMA electrical field patterns (2007) Simulated were three antenna designs 1) Vertex (built by VertexRSI): space framework struts 2) AEM (built by Thales Alenia, European Industrial Engineering, and MT Aerospace): elliptical struts with cladding 3) ideal antenna: no struts MELCO 12m and 7m AEM Vertex 8

9 b) enable use of response images Best source of ALMA responses so far: The TICRA simulations of ALMA electrical field patterns (2007) Simulated were three antenna designs 1) Vertex (built by VertexRSI): space framework struts (also good approx. of Melco) 2) AEM (built by Thales Alenia, European Industrial Engineering, and MT Aerospace): elliptical struts with cladding 3) ideal antenna: no struts (not used here) MELCO 12m and 7m AEM Vertex 9

10 b) enable use of response images The TICRA simulations of ALMA electrical field patterns Detailed front end geometry with secondary mirror taken into account where available: Bands 1, 5, 8: design still preliminary in 2007, only Gaussian beam simulated Bands 3, 4, 6, 7, 9: complete simulation at 3 or 4 frequencies within band Bands 2, 10: design incomplete in 2007, not simulated Electric Field Patterns (co- and cross-polar for both X and Y) in GRASP format Examples of front end geometries: Band 3 (left), Band 7 (right) 10

11 b) enable use of response images The TICRA simulations of ALMA electrical field patterns Simulated Electric Field Patterns (copolar field, X) for bands 1 and 3 to 9 (Vertex Struts) Band 1 (31 GHz) Band 3 (100 GHz) Band 4 (144 GHz) Band 5 (187 GHz) Band 6 (243 GHz) Band 7 (324 GHz) Band 8 (500 GHz) Band 9 (720 GHz) Simulated Electric Field Patterns (crosspolar field, X) for bands 3, 4, 6, 7, 9 (Vertex Struts) Band 3 (100 GHz) Band 4 (144 GHz) Band 6 (243 GHz) Band 7 (324 GHz) Band 9 (720 GHz) Normalised magnitude of E-field 11

12 b) enable use of response images The TICRA GRASP format response images converted to Voltage Pattens in CASA Image format. CASA response image format agreed with NRAO imaging team: - Four-plane (in the Stokes dimension) complex-valued image containing the voltage pattern for the parallel hands (XX and YY) and the cross-hands (XY and YX). - Directional coordinate system of type AZELGEO centered on (0.,0.). - Reference frequency of the responses stored by creating a degenerate spectral axis centered on that frequency. CASA 3.4 and 4.0: vpmanager part of CASA but not yet connected to CASA imaging. Cannot use the new ALMA PB library in normal imaging. But a patch was made to permit at least post-deconvolution correction with a single PB (homogeneous array case). With this patch, more tests were carried out. 12

13 Tests on ALMA data Second test dataset: January 2012, 4 SPWs in Band 3, 19 antennas (14 DV, 3 PM, 2 DA) 61 point mosaic on Quasar samples 9 different offsets * * * * * Image each pointing individually, * * * * * * re-center resulting images to the nominal Quasar position, * * * * * * * * * * * * * * * then superimpose using individual masks easily understandable plots like this: * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 13

14 Tests on ALMA data ALMA Early Science, 99.9 GHz, hour angle = -1 h, analysed with CASA 3.3 default ALMA PB correction 14

15 Tests on ALMA data ALMA Early Science, 99.9 GHz, hour angle = -1 h, analysed with CASA 3.3 default ALMA PB correction rel.flux arcs 15

16 Tests on ALMA data ALMA Early Science, 99.9 GHz, hour angle = -1 h, analysed with CASA 3.3 without any PB correction rel.flux arcs 16

17 Tests on ALMA data ALMA Early Science, 99.9 GHz, hour angle = -1 h, analysed with CASA 3.4 default PB corr. (scaled Airy disk) rel.flux arcs 17

18 Tests on ALMA data ALMA Early Science, 99.9 GHz, hour angle = -1 h, analysed with CASA 3.4 using TICRA sim. PB image rel.flux No systematic deviations out to at least 2 PB radii arcs 18

19 Tests on ALMA data ALMA Early Science, 99.9 GHz, hour angle = -1 h, analysed with CASA 3.4 using TICRA sim. PB image rel.flux Remaining scatter mostly due to noise arcs 19

20 Tests on ALMA data - status Tests ongoing. - See consistent results at other freqs. in Band 3. - Frequency scaling seems to work (have between 2 and 3 responses per band, use scaling to interpolate) - Analysis of Band 6 data under way. More data forthcoming - For tests of parallactic angle issues (field rotation) need brighter source at larger offsets - Crosshands still need to be investigated when ALMA data with full polarisation become available. - Following slides: Some examples of the simulated responses 20

21 Simulated Voltage Pattern Examples Vertex XX Voltage Pattern (Band 3, 100 GHz) 21

22 Simulated Voltage Pattern Examples AEM XX Voltage Pattern (Band 3, 100 GHz) 22

23 Simulated Voltage Pattern Examples AEM XX Voltage Pattern (Band 7, 324 GHz) 23

24 Simulated Voltage Pattern Examples AEM XX Voltage Pattern (Band 9, 661 GHz) 24

25 additional ray-tracing simulations Finally: working on alternative ray-tracing simulation (based on code by W. Brisken) of beams not available from TICRA (via aperture illumination pattern) (e.g. 7m beams, Bands 2, 10, and 11, and astigmatic reflectors in all bands) Up to first null, results look promissing; good agreement with TICRA simulation Prototype is available in CASA 4.0. Vertex XX 100 GHz TICRA sim. Vertex XX 100 GHz ray-traced 25

26 additional ray-tracing simulations Finally: working on alternative ray-tracing simulation (based on code by W. Brisken) of beams not available from TICRA (via aperture illumination pattern) (e.g. 7m beams, Bands 2, 10, and 11, and astigmatic reflectors in all bands) Examples of simulated ALMA 12 m beams with astigmatism: Vertex XX 100 GHz Z5 = 0.05 Vertex XX 100 GHz Z6 =